Introduction

There has been an alarming increase in carbon emissions in the last century, the main driver of climate change (Rockström et al. 2017). In 2016, this resulted in the Paris Agreement which aimed to limit temperature rise to 2 °C above pre-industrial levels with an optimal target of 1.5 °C. In order to achieve this, a severe reduction in global carbon emissions in the next thirty years is required, this is referred to as decarbonisation (or deep decarbonisation) (Åhman et al. 2017).

The Paris Agreement acted as a catalyst for investment in low carbon and cleaner technologies (Ivanova 2016). Various governments introduced several policy tools to encourage innovation and development in the field of sustainability, to allow economies to transition to low carbon alternatives and achieve decarbonisation (Roberts et al. 2018). Encouraging sustainable innovation represents a substantial challenge (Grubb et al. 2021); this is an especially daunting task for industries that are characterised as “hard to decarbonise”, such as iron and steel (Boldrini et al. 2024), cement (Habert et al. 2020), and chemicals (Griffiths et al. 2021). A further, more wide-ranging area that represents a substantial decarbonisation challenge is heating, which has been recognised as a significant global challenge (Abbasi et al. 2021).

Several policy instruments have been employed to promote sustainability and to foster green or sustainable innovation (Kern et al. 2019); amongst them is the funding of university led research projects (Calignano and Trippl 2020). Funded university led projects are used to facilitate university–industry collaboration, which are viewed by both parties as relatively low risk agreements in the development of green technologies (Dall-Orsoletta et al. 2022).

The development of complex knowledge, such as decarbonisation technologies, requires inputs from collaborators and agents (Jones 2009). Therefore collaboration, and encouraging collaboration, both at the national and international level is salient to the challenge of decarbonisation, especially as collaboration networks are critical in the development of new technologies and the cross pollination of ideas (Fadeeva 2005; Miguélez and Moreno 2015). Collaboration and knowledge exchange networks have been found to be of particular importance in the production of clean technologies (Boezeman and de Coninck 2018; Marra et al. 2020).

Extant work on knowledge exchange often draws on the Triple Helix Model (Etzkowitz and Leydesdorff 1995; Leydesdorff 2000; Galvao et al. 2019), which emphasises the need for dynamic interactions between university, industry, and government agents during innovation processes. The triple helix theory identifies that universities are central to knowledge production, and act as engines for the development of innovation capabilities at various levels (both regional and national) (Yusuf and Nabeshima 2006; Coenen 2007). The triple helix model has also been applied to understand eco-innovation (Yang et al. 2012). For example, Lerman et al. (2021) apply the triple helix model to the case of Germany to examine the different roles that universities, industry actors and government agents play in renewable energy systems. They find that the three actor types play varying roles in relation to renewable energy policy and highlight that the contribution of universities impacts knowledge generation at the regional level.

This study focuses on research funding as the policy tool to tackle climate change and transition to a low carbon economy. Schulze (2021) notes that domestic policy is vital to climate change and outlines hard and soft policies that can be applied to tackle climate change. Soft policy instruments include subsidies and research grants, whilst hard policy instruments include taxes and regulation. Governments often pursue soft policies in the first instance, as hard policies have historically not been viewed as “vote winners”, while their potential to distort markets makes them instruments of last resort. Table 1 provides an overview of selected studies that examine patterns for funded projects, with a focus on funded projects tackling climate change. A number of patterns can be observed from Table 1, more specifically, a tendency for the uneven distribution of funding for climate change research. Whilst the studies presented in Table 1 present aggregate overview of funding patterns for climate change research, this study intends to examine funded projects linked to industrial decarbonisation in further detail, by examining the collaboration networks that emerge from funded projects.

Table 1 Summary of selected works examining funded project patterns
Full size table

Decarbonisation and transitioning to low carbon technologies would allow a country to address three of the UN’s Sustainable Development Goals (SDGs), in particular Goal 7 (Affordable and Clean Energy), Goal 12 (Responsible Consumption and Production) and Goal 13 (Climate Change) (Voitko et al. 2021). Therefore, national decarbonisation initiatives represent key a step towards achieving important SDGs and long term sustainability (Anderson and Gough 2021). Kulkarni et al. (2022) examine the investment needs to achieve the SDGs; they find that the SDG with the largest investment gap is for SDG 13 (Climate Change).

An important step for many countries in achieving the SDGs and transitioning to a low carbon economy is the adoption of circular economy practices and increased utilisation of renewable energy sources (Ranjbari et al. 2022a). In last decade, investment in biofuels and biomass has gained momentum, as it represents a key renewable energy source and a substitute for fossil fuels (Banja et al. 2019; Tabatabaei et al. 2020). Panoutsou et al. (2021) note that the use of biofuels represents a core aspect of the strategy to decarbonise the European transport sector; yet they note that biofuel production costs remain higher than that of fossil fuels, and that there is a need for significant innovation and technological development (Li et al. 2015). Ranjbari et al. (2022b) conduct a systematic bibliometric analysis of the biomass potentials and identify a number of key funders in this research area. They note that the European Commission is a dominant funder, followed by UK research councils and the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (a Brazilian government agency).

University knowledge exchange and technological development, especially in the area of green and sustainable technology, can be viewed as universities delivering on their third mission beyond teaching and research: contribution to society (Trencher et al. 2014). Whilst there is substantial literature examining the determinants of firms participating in the development of eco-innovations and clean energy (Kesidou and Demirel 2012; Nesta et al. 2014), the activities of universities are often not examined in detail. Yet government funding into clean energy innovation is a major policy tool and is often implemented through the allocation of substantial research funds to universities on specific projects. Quatraro and Scandura (2019) acknowledge this in their study and find that participation of academic inventors in patenting activity has positive effects on the generation of green technology due to increased interdisciplinary insights. Fabrizi et al. (2018) also note that firm participation in green European research networks (in particular with universities and public research centres) positively affects environmental innovation.

Several studies have mapped collaboration networks arising from funded projects and investigated the implications for knowledge generation and innovation. A selection of these studies are summarised in Table 2, which indicates the study, the context, the network under analysis and the key findings. Table 2 indicates some of the key findings from this line of research, such as the tendency for these collaborations network to be characterised by hierarchical network structures, such as the presence of “rich clubs”. These studies tend to focus on all projects funded by certain institutions, this paper aims to contribute to an area that has received less attention: drawing on project funding data to map the research landscape for a highly important area—industrial decarbonisation.

Table 2 Overview of key studies examining organisational collaboration networks from funded project
Full size table

A further area that has received substantial interest in recent years is the relationship between economic variables and environmental quality (Kaika and Zervas 2013; Ardakani and Seyedaliakbar 2019). One of the most frequently used frameworks to understand patterns of carbon emissions is the EKC (Environmental Kuznets Curve) hypothesis, which was first developed by Grossman and Krueger (1995). The EKC hypothesis aims to explain the link between environmental quality and per capita income. The EKC hypothesis states that when a country has a low-income level (low GDP per capita), the environmental pollution (or environmental degradation) increases with increases in GDP per capita. When the income level reaches a certain point, the environmental pollution decreases (the environmental quality improves) with increases in GDP per capita. Therefore, the EKC hypothesis proposes an inverted U-shaped curve between environmental quality and GDP per capita.

The extant literature on the EKC hypothesis often provides empirical investigations to test the EKC hypothesis for various regions and countries (Uchiyama 2016; Sarkodie and Strezov 2019). There have been mixed levels of support for the EKC hypothesis. For instance, Bölük and Mert (2014) find no support for the EKC hypothesis for 16 European countries over the period from 1990 to 2008. However, Dogan and Seker (2016) provide empirical support for the EKC hypothesis for the European Union in their analysis of carbon emissions over the period 1980–2012; confirming the findings of Acaravci and Ozturk (2010). In addition to carbon emissions, a number of other sustainability related variables have been found to have a relationship with GDP per capita. For instance, Rajaeifar et al. (2017) note that municipal solid waste generation is coupled with economic growth.

This study considers the case of the UK (drawing on UK research council funding data) and Europe more widely (drawing on Horizon 2020 data). This study examines the funded projects related to industrial decarbonisation and aims to better understand the knowledge exchange networks that arise from these projects. This study aims to examine how a country’s position in the collaboration network arising from projects on industrial decarbonisation in Horizon 2020 impacts a key environmental indicator. A number of studies have examined the impact of innovation and technology investments on country CO2 emissions and other environmental indicators (Álvarez-Herránz et al. 2017; Ganda 2019). This study proposes to build on these, by not only examining aggregated funding levels, but also the position a country holds in a key knowledge exchange network. This paper addresses the following questions about the research landscape of industrial decarbonisation:

  • RQ1: Who are the key institutions involved in research council funded projects linked to industrial decarbonisation in the UK?

  • RQ2: What is the structure of the inter-organisational collaboration network arising from these collaborative arrangements?

  • RQ3: What are the features of collaboration networks (at both the country and organisational level) arising from collaborative arrangements on decarbonisation related projects funded under Horizon 2020?

  • RQ4: Does country level participation in the collaboration network arising from participation on decarbonisation related projects funded under Horizon 2020 have an impact on environmental indicators?

  • RQ5: Do find support for the EKC hypothesis when examining the link between country level attributes and environmental indicators?

Data and Methods

This paper draws chiefly on two data sources when investigating the research landscape for the UK and Europe. For the UK, this paper utilises the data on projects funded by UK research councils, that has been extracted (programmatically) from Gateway to Research (GtR); this dataset has been used to examine collaboration networks in the sciences (Smith et al. 2022; Sarabi and Smith 2023) and arts and humanities (Williams et al. 2017). GtR provides information on a variety of different types of projects funded by UK research councils, such as studentships, research grants, knowledge exchange programmes, yet this study focuses on research grants. The information on research grants provided by GtR includes the organisational collaborative arrangements, funding level, start and end dates, along with project output data.

In addition to the UK case, this study also examines projects funded under Horizon 2020. Horizon 2020 is the EU Framework Programme that funds research over the period of 2014–2020 and is recognised as key financial instrument for research and innovation globally (Veugelers et al. 2015; Kim and Yoo 2019). It has also been noted that Horizon 2020 represents an important funding body for sustainability issues (in a range of disciplines) (Foulds and Christensen 2016; Pollex and Lenschow 2018). Horizon 2020 data has been utilised to examine knowledge exchange and research patterns in a number of disciplines (Enger 2018; Enger and Gulbrandsen 2020).

The projects examined in the GtR and Horizon 2020 data are those with any of the keywords given in Table 3 in either the title or abstract. These allow for the examination of projects relevant to industrial decarbonisation.

Table 3 Search terms
Full size table

To address the first and second research questions posed by this paper, this study draws on social network analysis to examine the key organisations and countries in the decarbonisation research landscape. This study examines the inter-organisational collaboration networks arising from collaborative arrangements of UK research council and Horizon 2020 funded projects. The inter-organisational networks consist of organisations as actors (including both universities and non-academic organisations), who are linked if they work on the same project. The network ties are valued: the value of the tie represents the total value of funded projects that the connected organisations collaborate on together. Inter-organisational networks have been frequently utilised to investigate how knowledge exchange assists in developing innovative activities (Olmos-Peñuela et al. 2014; He et al. 2021).

Several centrality metrics are used to identify organisations holding key potions in these collaboration networks, more specifically, weighted degree centrality, degree centrality, betweenness centrality and eigenvector centrality. Degree centrality captures the number of collaborative partners an organisation has (Freeman 1978), and weighted degree centrality is the number of partners weighted by the intensity (or value) of the tie (Opsahl et al. 2010). Betweenness centrality captures the brokerage potential of an actor in the network, more formally it is the number of times an actor sits on the shortest path between two other actors in the network (Freeman 1977). An actor with a high betweenness centrality may act as an information broker in collaboration or knowledge exchange networks, and therefore may hold a strategically important position. Eigenvector centrality is often viewed as a measure of status or prestige, as an actor with high eigenvector centrality is connected to other well connected actors in the network (Bonacich 2007).

To address the third research question posed by this paper, the country-to-county network arising from the Horizon 2020 data is analysed. The country level network aggregates the Horizon 2020 data to the country level, where nodes in the network are countries and they are linked if organisations or universities from one country have collaborated with organisations or universities from another.

An OLS regression model is employed to examine the relationship between a key environmental indicator (coal share in 2021) and country level activity (captured by the various centrality measures) in industrial decarbonisation research (over the period 2014–2020). Country level activity in decarbonisation research is captured by calculating a country’s centrality in the country collaboration network based on the Horizon 2020 data. The outcome variable used in this study is for 2021 and is the share of electricity consumption that comes from coal. Coal use is one of the largest sources of CO2 emissions in the global economy (Glomsrød and Taoyuan 2005; Li et al. 2020). This variable is also used in to track country progress towards the Sustainable Development Goals (SDGs) (Ritchie and Roser 2020).

Within the model specification, a number of other country level variables as dependent variables. These include GDP per capita and GDP per capita squared, in order to test the EKC hypothesis (for the share of electricity consumption that comes from coal) and address the fifth research question posed by this paper. Trade openness is also controlled for in the model specification. A separate model is implemented for each centrality measure, as extant work has highlighted centrality measures are often correlated (Schoch et al. 2017).

Results and Discussion

The results of analysis of the UK research landscape on projects related to decarbonisation will be presented first, this will be followed by an analysis of the European case (as captured by Horizon 2020 data). The results and discussion section will conclude with an analysis of the activity of countries participating in EU decarbonisation projects and share of electricity consumption that comes from coal.

UK Decarbonisation Research Landscape

The search produced 435 projects related to industrial decarbonisation that were funded by UK research councils. There were a number of key outputs produced by these 435 projects, including 5995 published journal articles (on average around 14 published articles per project). SCImago journal rank (SJR) indicator (SCImago 2020) are used to examine the average quality of the journal articles produced by the projects linked to industrial decarbonisation. The SJR is a measure of journal prestige (González-Pereira et al. 2010), that takes into account both the number of citations a journal receives, and the prestige level of the citing journal (Falagas et al. 2008). On average, the journal quality of these projects was 2.22 SJR, which reflects high quality (Q1) published work. Additionally, 113 of these projects produced policy related documents.

In the UK there are several research councils that fund projects in a variety of disciplines. Figure 1 indicates the level of funding allocated by each UK research council to industrial decarbonisation related projects. The BBSRC is the Biotechnology & Biological Sciences Research Council (BBSRC), and it is one of the largest UK research councils (by number of projects funded), and the discipline areas they cover includes green energy, food security issues, industrial biotechnology, and pharmaceuticals. Given the topic areas it covers, it is somewhat surprising that they are not a more active funder of industrial decarbonisation.

Fig. 1
figure 1

Total funding awarded by UK research councils for industrial decarbonisation related projects (2011–2020)

Full size image

The research council that allocated the most funding to industrial decarbonisation projects is the EPSRC, the Engineering & Physical Sciences Research Council (EPSRC). The EPSRC funds a number of disciplines in the sciences, such as healthcare technologies, structural engineering, manufacturing, mathematics, advanced materials, and chemistry. They are one of the largest public funders in the UK for research related to innovation activities (Owen and Goldberg 2010). Given the engineering focus of this research council, it is not unexpected that it is a major funder of industrial decarbonisation related projects, given that decarbonisation is often linked to engineering and industrial capacity (Jones 2021).

The Natural Environment Research Council (NERC) is also a funder of industrial decarbonisation related projects. Figure 1 indicates it is the second largest funded of these projects. Other research councils shown in Fig. 1 which are not major funders of industrial decarbonisation, due to discipline areas that do not strongly align with decarbonisation are the Arts & Humanities Research Council (AHRC), the Medical Research Council (MRC) and the Economic & Social Research Council (ESRC). The Science and Technology Facilities Council (STFC) is also a minor funder of decarbonisation research.

The results presented in Fig. 1 are generally in line with the findings of Sovacool et al. (2022). In their analysis of a funded projects linked to climate change, spanning a range of countries, they note that with the exception of Horizon 2020, UK research councils are the dominant funders of energy and climate research, with the top funders as the ESPRC, ESRC and NERC.

Figure 2 indicates the level of funding awarded by each research council to industrial decarbonisation projects over time. A steady increase in the amount allocated by the EPSRC to industrial decarbonisation related research from 2014 to 2018 (with a sharp increase in 2018) can be observed. However, less funding has been awarded since 2018, with a drop in 2019 and 2020. This perhaps reflects that other areas, such as COVID-19, became a priority over decarbonisation research. However, many argue that compared to the permanent havoc that climate change threatens, COVID-19 is a smaller challenge, and should not have caused the diversion of funds (Bloomfield and Steward 2020). The drop in funding for climate research has also been observed by AbdulRafiu et al. (2022) in their examination of funding patterns across countries. They find that a sharp increase in climate change funding from 2014 to 2018, yet this is more pronounced in energy systems research compared to industrial decarbonisation research. They also observe a reduction in funded projects after 2018. From Fig. 2 it can be observed that the level of funding awarded by the other UK research councils has remained relatively constant over this time period.

Fig. 2
figure 2

Level of funding awarded per year by UK research councils for industrial decarbonisation related projects (2011–2020)

Full size image

Figure 3 is a word cloud of the top terms in each of the UK research council funded industrial decarbonisation related project’s title and abstract. Figure 3 indicates that (with the exception of somewhat trivial keywords, such as energy and carbon), heat and (to a lesser extent) storage are priority topics in research projects funded by UK research councils (Janes-Bassett et al. 2021). One surprising omission is biomass (or biofuel), given its important as a renewable energy source (Ranjbari et al. 2022a), and that it has been viewed as an option for decarbonisation by policy makers (Levidow and Papaioannou 2016). This perhaps indicates the need for UK research councils to fund more targeted projects, rather than those operating in the broad area of energy research.

Fig. 3
figure 3

Word cloud of top terms in titles and abstract of UK research council funded projects linked to industrial decarbonisation (2011–2020)

Full size image

Figure 4 is a tree map indicating the top organisations by amount of funding they receive (in the capacity of lead organisation) from UK research councils for industrial decarbonisation related projects. Figure 4 indicates that many of these universities are Russell Group universities. The Russell Group are a set of twenty four universities with a focus on research intensive activities and Russell Group membership is frequently used as a proxy for elite status, especially in extant work investigating university cooperation for innovation purposes (Hewitt-Dundas 2013; Guerrero et al. 2015). Out of the 435 projects linked to industrial decarbonisation funded by UK research councils, 283 (65%) had a Russell Group university as the lead organisation.

Fig. 4
figure 4

Tree map of top lead organisations for industrial decarbonisation projects funded by UK research councils. Size & colour of the title refers to total level of funding awarded to each organisation

Full size image

A further point to highlight is that a number of London based institutions are recipients of high levels of funding (UCL and Imperial College London), and Welsh institutions are also playing prominent roles in the UK research landscape (especially Swansea University). A number of studies have focused on the Welsh case when examining decarbonisation patterns, notably the case of the Port Talbot Steelworks in South Wales (Williams et al. 2021).

Figure 5 gives the inter-organisational network arising from collaborative arrangements on these decarbonisation projects. Organisations are linked if they have worked on the same project together, and the ties are weighted by the total value of the projects they have collaborated on together. In Fig. 5, the blue squares are non-academic organisations, the green triangles are non-UK universities, and the red circles are UK universities. On average each project consisted of around 6 or 7 organisations (both academic and non-academic). On average, around 53% of the collaborating organisations working on a project were non-academic organisations. In Fig. 5 a core-periphery structure can be observed with UK universities at the very centre.

Fig. 5
figure 5

Inter-Organisational Collaboration Network (UK) Blue squares are non-academic organisations, green triangles are non-UK universities and red circles are UK universities

Full size image

Table 4 provides some descriptive statistics for the UK inter-organisational collaboration network. Network size captures the number of organisations in the network. Density is a measure of network connectivity, more specifically it is the ratio of observed ties to all possible ties in a network (Wasserman and Faust 1994). Degree centralisation captures the extent to which the network reflects a star like structure, where degree centrality is concentrated in the handful of few actors in the network (Sinclair 2011; Borgatti et al. 2018). The clustering coefficient is a measure of network cohesion and represents the average of the densities of the neighbourhoods of all actors in the network (Watts and Strogatz 1998). The correlation between degree centrality of connected actors in a network is referred to as degree assortativity (Newman 2002). Degree assortativity scores can be between − 1 and 1, where a positive score would suggest that well-connected actors link to other well-connected actors. A negative score, on the other hand, would suggest that high degree actors are connected to many low degree actors, which is referred to as a pattern of disassortativity in the network and indicates hierarchical relations among actors.

Table 4 UK Inter-organisational network descriptive statistics
Full size table

A low network density is observed, pointing towards low levels of connectivity in this inter-organisational network. The degree centralisation score suggests that degree centrality is not necessarily concentrated in a small handful of actors in the network. The clustering coefficient is relatively high, suggesting that areas of the network characterised by level of high connectivity, such as in the core of the network, as observed in Fig. 5. The degree assortativity is positive, (yet only slightly). This points towards a tendency for well-connected actors to connected to other well-connected actors. The degree assortativity suggest the collaboration network is characterised by a so-called rich-club, supporting the findings of Ma et al. (2015). A rich club nodes with a high level of connections tend to connect to other well-connected nodes (Colizza et al. 2006).

Table 5 gives the top organisations by the various centrality measures for the inter-organisational collaboration network arising from collaborative arrangements on UK research council funded projects. Table 5 indicates that many of the universities holding central positions in this network are Russell Group universities. It is not necessarily surprising that research intensive institutions that receive the largest share of government funding are in prominent positions (O’Connell 2015). These results confirm the findings of Smith et al. (2022) for the case of industrial decarbonisation projects; they find that Russell Group members are more active in forming collaborative ties in research council funded projects (especially in EPSRC funded projects). In addition to elite universities, Table 5 indicates that there are a number of other institutions holding central positions in this inter-organisational collaboration network.

Table 5 Top ten organisation by centrality scores calculated on the intra-organisational collaboration network (UK)
Full size table

Weighted degree centrality captures the number of collaborating partners an organisation has weighted by the value of projects they collaborate on. Table 5 indicates a number of key non-academic actors with a high weighted degree centrality score. More specifically, the key non-academic institutions holding central positions are the Welsh Government, National Grid UK, Tata Steel, Arup, and the Department of Transport. This suggests that, in line with predictions by the triple helix model, University–Industry–Government ties are critical for knowledge exchange in the area of industrial decarbonisation. The presence of the National Grid in the top most central actors is perhaps unsurprising given its embeddedness in the wider UK energy system (Dodds and McDowall 2013; Logan et al. 2022). Arup is a multinational firm that provides services linked to the built environment and has been involved in research linked to decarbonising the construction industry (Sweet 2019). The presence of the Welsh Government and Tata Steel as highly central actors reflects the funding provided by UK research councils to support an industry that is “hard to decarbonise”—the steel industry, given that Port Talbot Steelworks in South Wales is home to Tata Steel’s operations in Europe (Williams et al. 2021).

When examining degree centrality, that is just a count of the number of collaborating partners (not weighted by project value), Table 5 indicates the prominent role of elite universities connecting to a large number of other organisations in the UK decarbonisation research landscape. Betweenness centrality captures the potential knowledge brokering roles of organisations in this system. When examining top ranked organisations, in addition to elite, Russell Group universities, key science centres (and action groups) also hold brokering positions, indicating key roles in knowledge generation in the area of industrial decarbonisation. Eigenvector centrality, often viewed as a measure of status and prominence in the network as it detects well connected actors linked to other well-connected actors, provides similar results to weighted degree centrality. The key institutions holding prominent positions in the UK research system for industrial decarbonisation are elite universities, critical industry actors (such as National Grid and Tata Steel), and big science centres. Ferrer-Serrano et al. (2021) observe similar patterns in their analysis of European collaboration networks, noting that although larger science centres do not receive high level of funding, they play salient role as connectors, driving innovation activities in collaboration systems.

Horizon 2020 Decarbonisation Research Landscape

In this section the research landscape for decarbonisation related projects funded under Horizon 2020 is examined. Applying the search terms presented in the data and methods section to Horizon 2020 projects returns 239 projects related to industrial decarbonisation. This is substantial less than the number of projects funded by UK research councils, reflecting the UK’s role as a major funder of decarbonisation research (Sovacool et al. 2023).

Figure 6 indicates the top twenty countries by the level of funding received for Horizon 2020 industrial decarbonisation related projects. Figure 6 indicates that the countries receiving the lion’s share of income on decarbonisation projects are the UK, Netherlands, Italy Spain, and France, indicating funds are often allocated to organisations belonging to Western European nations. Given the varying patterns of inequality of energy consumption and carbon emissions within Europe (Bianco et al. 2019, 2021), this suggests that future European funding initiatives should seek to allocate funding to a wider variety of nations, rather than the wealthiest economies in Europe.

Fig. 6
figure 6

Top countries by level of funding allocated for industrial decarbonisation related projects under Horizon 2020

Full size image

Figure 7 indicates the level of funding awarded by Horizon 2020 to decarbonisation projects each year from 2014 to 2020. Figure 7 indicates two spikes in funding in 2016 and 2019. The spike in funding in 2016 might reflect the increase in interest in decarbonisation tiggered by COP21 and the associated Paris agreement.

Fig. 7
figure 7

Level of funding awarded by Horizon 2020 for industrial decarbonisation over time

Full size image

Figure 8 is a word cloud of the top terms in each of the project title and abstract for industrial decarbonisation related projects. The word cloud is similar to the one presented for the UK case of the funding landscape, observed in Fig. 3. However, compared to the UK case, there is a wider range of topics funded under Horizon 2020 associated to “hard to decarbonise” fields such as cement. The frequency of storage and heat is comparable to the UK case, yet cement is more visible. While iron, steel and chemicals (other “hard to decarbonise” sectors) do not hold prominent positions on the word clouds for either the UK or Horizon 2020 case, perhaps reflecting research areas that could be better funded. Biomass (or biofuels) is not part of this word cloud, similar to the UK case; again this is a surprising result, given that this is viewed as a attractive renewable energy source by EU policy makers (Balat 2007).

Fig. 8
figure 8

Word cloud of top terms in titles & abstracts of Horizon 2020 funded decarbonisation projects

Full size image

Figure 9 gives the inter-organisational collaboration network arising from collaborative arrangements on industrial decarbonisation projects funded under Horizon 2020. Similar to Fig. 5 for the UK case, in Fig. 9 organisations are linked if they have worked on the same projects together, and the ties are weighted by the total value of the projects they have collaborated on together. In Fig. 9, the blue squares are non-academic organisations, and the red circles are universities. An important difference between the UK case and Horizon 2020 is that universities are not dominating the core of the network to the same extent in Fig. 9. This is in line with the findings of Ferrer-Serrano et al. (2021), who note that private firms are beneficiaries of high levels of public funding and therefore play a key role in the European research landscape.

Fig. 9
figure 9

Inter-Organisational Collaboration Network (Horizon 2020). Blue squares are non-academic organisations and red circles are universities

Full size image

Table 6 provides some descriptive statistics for the Horizon 2020 inter-organisational collaboration network, as visualised in Fig. 9. Many of the features of the inter-organisational network arising from Horizon 2020 projects are similar to the UK case, as presented in Table 4. The number of organisations, level of connectivity and clustering (as captured by size, density, and clustering coefficient respectively) is comparable (yet slightly lower) to the UK case. Degree centralisation is lower in the case of Horizon 2020, when compared to the UK case, suggesting that degree centrality is more evenly distributed than in the UK research landscape for industrial decarbonisation. Degree assortativity is negative (yet only slightly), contrasting to the UK case. This suggests a (weak) tendency for well-connected actors (with a high degree centrality) to be connected to peripheral actors (with a low degree centrality). This contrasts to the findings of Ferrer-Serrano et al. (2021), as it does not suggest there is a rich club for the collaboration network for industrial decarbonisation projects. This suggests that the rich club they detect in the full Horizon 2020 may be driven by other subject areas.

Table 6 Horizon 2020 inter-organisational network descriptive statistics
Full size table

Horizon 2020: Country Level Analysis

In this section the country-to-country network arising from Horizon 2020 collaboration arrangements is examined. Figure 10 presents the visualisation of this country-to-country network. The node size and label size represent weighted degree centrality (i.e. the number of ties, weighted by the intensity of the interaction). Extant research has shown that countries tend to collaborate with those that have a reputation for research excellence, who share a language, have similar institutional features and a record of previous collaborations (Vicente et al. 2008; Breschi and Lissoni 2009; Hoekman 2012; Garrone et al. 2014). Balland et al. (2019) provide an examination of projects funded by the EU from 2003 to 2017 and examine the country level network arising from project collaborations. They find that EU-15 nations tend to hold central, dominant positions in these networks, with a high number of collaborative ties. They note that larger countries tend to hold more central positions in the collaborative network. In the case of Fig. 10, a core-periphery structure is observed, with wealthy economies connected to each other at the centre. These results are in line with previous empirical work examining country collaboration networks.

Fig. 10
figure 10

Country collaboration network (Horizon 2020). Nodes size and label size represent weighted degree centrality

Full size image

Table 7 gives the descriptive statistics for the country level network. The network has a high level of connectivity, and a high clustering coefficient, which suggests that the network is characterised by areas of high local density. This is observed in Fig. 10, where the centre of the network is densely connected. Degree centralisation is higher than what was observed for the inter-organisational collaboration networks; this suggests that degree centrality is concentrated in a few actors at the very centre of the network. The negative degree assortativity result points towards a pattern of disassortativity in the network, with high degree actors connected to many low degree actors.

Table 7 Country network descriptive statistics
Full size table

Table 8 provides the top five most central countries in this network. In terms of weighted degree, binary degree and eigenvector centrality, the top countries are typically the wealthiest nations. In the case of degree centrality, the role of Germany is not necessarily surprising, given its strengths in the development of green technologies (Kim et al. 2023). Greece also holds a prominent position in terms of degree centrality. This might reflect support provided under Horizon 2020 for research dedicated to accelerating the delignitization of Greece's electricity generation (Roinioti and Koroneos 2019; Karamaneas et al. 2023). This is further demonstrated in the case of weighted degree centrality and eigenvector centrality. However, there are notable differences when considering betweenness centrality, where it is not only the rich nations at the core of the network that play knowledge brokering roles. One surprising result in the betweenness centrality rankings is the presence of Bulgaria, which is actually known for its reliance on fossil fuels and rather weak performance regarding decarbonisation progress (Spencer et al. 2017; Primova 2020; Ciołek et al. 2022). This potentially suggests that Bulgaria is not benefitting from its brokerage position, highlighting the need for funders to be aware of the risks of misallocating climate change funding (Overland and Sovacool 2020), especially if projects are not producing meaningful decarbonisation outcomes. Denmark has also been identified as a key broker, which reflects its strong record on climate change. Sovacool et al. (2018) note that Scandinavian countries have generally made good progress on the decarbonisation of electricity, heat, and buildings, yet lag behind on the decarbonisation of transport and mobility.

Table 8 Top five countries by centrality
Full size table

Table 9 presents the regression results for a set of models with only the centrality metrics, and another set with the country level controls. When the country level controls are included, the various centrality metrics, reflecting embeddedness in the decarbonisation collaboration network and activity in this field of research, are not associated with coal share. The results presented in Table 9 provide does not provide support for the EKC hypothesis, as there is a negative and significant GDP per capita, and a positive and significant GDP per capita squared; this points towards a U-shaped relationship between market affluence and coal share. These results are in line with the work of Dogan and Inglesi-Lotz (2020); they identify a U-shaped relationship between market affluence and CO2 for a set of European countries for the period 1980–2014. The results indicate no impact of trade openness on coal share, in line with previous work that has found no effect of trade openness on environmental indicators (Kearsley and Riddel 2010).

Table 9 Regression results
Full size table

However, when only the centrality metrics are included in the model specification, some statistically significant results are observed (yet only at the 0.05 level). One potential explanation is that the most dominant countries in the Horizon 2020 collaboration network tend to be more affluent nations, with a higher GDP per capita, as observed in Fig. 10, and confirmed by the centrality results presented in Table 8. It is also important to highlight that when these controls are not included, betweenness centrality is not significant. In Table 8, the countries with high betweenness are not necessarily the richest (rather have lower GDP per capita levels). The results suggest that there is a link between country affluence and participation in the knowledge exchange network arising from Horizon 2020 funding for decarbonisation rather than to the share of electricity consumption that comes from coal. Extant work has noted that the allocation of EU funding (such as Horizon 2020) is often skewed, with high levels allocated to top universities in Europe (which tend to be located in more affluent countries) (Henriques et al. 2009; Enger and Castellacci 2016).

Conclusions and Prospects

This study made use of data from UK research councils and Horizon 2020 in order to map the research landscape for industrial decarbonisation. This study provided an exploratory analysis and drew on network analysis to examine the inter-organisational and country level collaboration networks arising from these funding sources. This study tackled five research questions.

The first research question asked who are the key institutions involved in research council funded projects linked to industrial decarbonisation in the UK? In the case of the UK, this study found that, in line with the triple helix model, University–Industry–Government links are critical elements of the research landscape on industrial decarbonisation, as indicted by network analysis. The results also indicated that whilst science centres and action groups are not associated with projects receiving the highest level of funding, they play key, strategic roles as knowledge brokers in the UK research system on industrial decarbonisation (as indicated by the betweenness centrality analysis). The key universities at the core of the inter-organisational network, and often receiving high level of funding were Russell Group universities, the set of elite, research intensive universities.

The second research question asked, what is the structure of the inter-organisational collaboration network arising from these collaborative arrangements (on UK research councils funded projects)? The results indicate that the inter-organisational collaboration network arising from collaborative arrangement on UK research council funded projects linked to industrial decarbonisation had a hierarchical structure, and a core-periphery structure.

The third research question focused on industrial decarbonisation projects funded under Horizon 2020 (in order to capture the wider European case), and asked what are the features of collaboration networks (at both the country and organisational level) arising from collaborative arrangements on decarbonisation related projects funded under Horizon 2020? The results indicated that many of the feature of the European industrial decarbonisation research landscape were similar to the UK case. The inter-organisational network (as depicted in Fig. 9) showed that private firms were critical to network structure.

Country level patterns were considered to address the fourth research question posed. This study inquired on whether country level participation in the collaboration network arising from decarbonisation related projects are associated to a country’s share of electricity consumption that comes from coal. This study found that countries at the core of this country level collaboration network tended to be more affluent nations. The results did not indicate a significant relationship between centrality in the collaboration network and share of electricity consumption that comes from coal, after controlling for key economic variables. Rather these results indicate network position is linked to affluence rather than use of coal. The fifth research question asked whether there was support for the EKC hypothesis when examining the link between the share of electricity consumption that comes from coal and market affluence. The regression results do not provide support for the EKC hypothesis in this setting.

This study assumes that structural advantage in collaboration networks reflects successful research strategies can be instructive for many actors on this research landscape. While a number of research institutions hold positions of privilege commensurate with their reputation, such as the UK Russell group universities, this dominance is not absolute. This is particularly evident in the analysis of country positions in the Horizon data. Network metrics suggest that countries like Bulgaria and Turkey have managed to gain network prominence, and most likely research dividends, by strategizing brokerage, as evident in their betweenness rank. In other words, there is evidence here that certain collaboration strategies are more effective than others.

Practical Implications

This study has a number of practical implications, in particular for funders of climate change research. Firstly, the analysis of the project descriptors for industrial decarbonisation projects in both the UK and Horizon 2020 settings indicate that whilst there has been some focus on the “hard to decarbonise” sectors, these issues were somewhat underrepresented. This suggests that there is a need to prioritise funding in these areas, perhaps shifting away from funding general energy research.

Whilst this study provides an examination of funded projects linked to industrial decarbonisation, future research should seek to examine the impact of these studies. However, the data provided by both the UK research councils and EU funders on impact is rather limited. In the case of the UK, this is often restricted to publications or whether it has been mentioned in a policy documents. For sustainability related projects, especially those tackling decarbonisation challenges, there is a need to embed advanced sustainability assessment tools in project impact reporting, this should include reporting the assessment findings as a salient project output. Sustainability assessment tools available include life cycle assessment (LCA) tools which are used to evaluate environmental impact and energy consumption of a product or process throughout its entire life cycle (He and Yu 2020). A further assessment technique is exergy analysis, which is valuable tool for sustainability research, as it allows to locate, quantify, and assess thermodynamic efficiencies (Amid et al. 2021; Soltanian et al. 2022). Exergoenviromental analysis has been successfully applied in bioenergy research (Aghbashlo et al. 2021), indicating its value in tackling climate change and decarbonisation challenges. However, Aghbashlo et al. (2022) note that there are limits to using a single sustainability assessment tool, rather a integrated approach should be taken, using multiple techniques, as this can provide more reliable and accurate results.

Limitations and Future Research

Limits to the work presented here represent fertile avenues for future research. Firstly, when analysing the link between country participation in a collaboration network and coal-share, we focused on collaborations arising from Horizon 2020 funded projects, yet as demonstrated by the UK case, there is substantial funding provided by other sources, often at the national (rather than European) level. A further avenue for future research is to examine the link between country position in the collaboration network and a wider range of scores on environmental indicators.

A further limitation of this study is that it has mapped the broad industrial decarbonisation landscape, yet industrial decarbonisation contains a number of subthemes and sectors. For example, a core theme is “hard to decarbonise” sectors, the rise of carbon capture technologies, along with renewable energy considerations (including biomass). Future research should seek to unpack the industrial decarbonisation landscape further, examining the areas within the field of industrial decarbonisation in more detail.

The lack of a relationship between the network position of a country and to a country’s share of electricity consumption that comes from coal could also indicate issues in the way that science influences policy (Radaelli 1995). More specifically how are scientific findings regarding decarbonisation leading to policies on electricity consumption that comes from coal? This indicates a need for further research on how new evidence and scientific knowledge on decarbonisation arising from government funded projects enters policy circles, and how this information is used in policy formulation.

The study only mapped the research landscape for industrial decarbonisation space, mapping existing collaborations; future work should seek to analyse and examine how these collaboration can be optimised. For instance, are there key actors with important resources for knowledge generation that only hold peripheral positions in collaboration networks.

This study is limited to the UK and Europe, therefore, omitting a number of key greenhouse gas emitters (such as the US and China) (Zheng et al. 2019). Future research should seek to map the research landscape in a wider range of nations (especially those with environmental degradation issues).